| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 80.0, | |
| "global_step": 2000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.3969079554080963, | |
| "eval_runtime": 2.9689, | |
| "eval_samples_per_second": 33.683, | |
| "eval_steps_per_second": 4.379, | |
| "step": 25 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 1.0, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.5, | |
| "eval_loss": 0.47094520926475525, | |
| "eval_runtime": 3.0206, | |
| "eval_samples_per_second": 33.106, | |
| "eval_steps_per_second": 4.304, | |
| "step": 50 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 2.0, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.42, | |
| "eval_loss": 0.33409440517425537, | |
| "eval_runtime": 3.0742, | |
| "eval_samples_per_second": 32.529, | |
| "eval_steps_per_second": 4.229, | |
| "step": 75 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 3.0, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.3011420965194702, | |
| "eval_runtime": 3.0993, | |
| "eval_samples_per_second": 32.265, | |
| "eval_steps_per_second": 4.194, | |
| "step": 100 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 4.0, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.36, | |
| "eval_loss": 0.3118648827075958, | |
| "eval_runtime": 3.1211, | |
| "eval_samples_per_second": 32.04, | |
| "eval_steps_per_second": 4.165, | |
| "step": 125 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 5.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.37, | |
| "eval_loss": 0.32973361015319824, | |
| "eval_runtime": 3.1414, | |
| "eval_samples_per_second": 31.833, | |
| "eval_steps_per_second": 4.138, | |
| "step": 150 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 6.0, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.53, | |
| "eval_loss": 0.29277390241622925, | |
| "eval_runtime": 3.1486, | |
| "eval_samples_per_second": 31.76, | |
| "eval_steps_per_second": 4.129, | |
| "step": 175 | |
| }, | |
| { | |
| "best_epoch": 0, | |
| "best_eval_accuracy": 0.6, | |
| "epoch": 7.0, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.3078657388687134, | |
| "eval_runtime": 3.1562, | |
| "eval_samples_per_second": 31.684, | |
| "eval_steps_per_second": 4.119, | |
| "step": 200 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 8.0, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.2875213921070099, | |
| "eval_runtime": 3.1568, | |
| "eval_samples_per_second": 31.677, | |
| "eval_steps_per_second": 4.118, | |
| "step": 225 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 9.0, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.2906056344509125, | |
| "eval_runtime": 3.1601, | |
| "eval_samples_per_second": 31.644, | |
| "eval_steps_per_second": 4.114, | |
| "step": 250 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 10.0, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2903980016708374, | |
| "eval_runtime": 3.1623, | |
| "eval_samples_per_second": 31.622, | |
| "eval_steps_per_second": 4.111, | |
| "step": 275 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 11.0, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_accuracy": 0.52, | |
| "eval_loss": 0.2946452796459198, | |
| "eval_runtime": 3.1658, | |
| "eval_samples_per_second": 31.588, | |
| "eval_steps_per_second": 4.106, | |
| "step": 300 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 12.0, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.294162392616272, | |
| "eval_runtime": 3.1659, | |
| "eval_samples_per_second": 31.587, | |
| "eval_steps_per_second": 4.106, | |
| "step": 325 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 13.0, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.29346808791160583, | |
| "eval_runtime": 3.1677, | |
| "eval_samples_per_second": 31.569, | |
| "eval_steps_per_second": 4.104, | |
| "step": 350 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 14.0, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.29134005308151245, | |
| "eval_runtime": 3.1688, | |
| "eval_samples_per_second": 31.558, | |
| "eval_steps_per_second": 4.103, | |
| "step": 375 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 15.0, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2886267304420471, | |
| "eval_runtime": 3.1675, | |
| "eval_samples_per_second": 31.571, | |
| "eval_steps_per_second": 4.104, | |
| "step": 400 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 16.0, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2899758815765381, | |
| "eval_runtime": 3.1773, | |
| "eval_samples_per_second": 31.473, | |
| "eval_steps_per_second": 4.091, | |
| "step": 425 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 17.0, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28738337755203247, | |
| "eval_runtime": 3.1788, | |
| "eval_samples_per_second": 31.459, | |
| "eval_steps_per_second": 4.09, | |
| "step": 450 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 18.0, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2909628748893738, | |
| "eval_runtime": 3.1848, | |
| "eval_samples_per_second": 31.399, | |
| "eval_steps_per_second": 4.082, | |
| "step": 475 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 19.0, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "learning_rate": 0.015, | |
| "loss": 0.6674, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_accuracy": 0.47, | |
| "eval_loss": 0.29309767484664917, | |
| "eval_runtime": 3.1809, | |
| "eval_samples_per_second": 31.437, | |
| "eval_steps_per_second": 4.087, | |
| "step": 500 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 20.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 21.0, | |
| "eval_accuracy": 0.51, | |
| "eval_loss": 0.2908664643764496, | |
| "eval_runtime": 3.1834, | |
| "eval_samples_per_second": 31.413, | |
| "eval_steps_per_second": 4.084, | |
| "step": 525 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 21.0, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 22.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28550270199775696, | |
| "eval_runtime": 3.1765, | |
| "eval_samples_per_second": 31.481, | |
| "eval_steps_per_second": 4.092, | |
| "step": 550 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 22.0, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 23.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.28806307911872864, | |
| "eval_runtime": 3.1841, | |
| "eval_samples_per_second": 31.406, | |
| "eval_steps_per_second": 4.083, | |
| "step": 575 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 23.0, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 24.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2877710461616516, | |
| "eval_runtime": 3.1789, | |
| "eval_samples_per_second": 31.457, | |
| "eval_steps_per_second": 4.089, | |
| "step": 600 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 24.0, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "eval_accuracy": 0.57, | |
| "eval_loss": 0.28737229108810425, | |
| "eval_runtime": 3.1767, | |
| "eval_samples_per_second": 31.479, | |
| "eval_steps_per_second": 4.092, | |
| "step": 625 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 25.0, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 26.0, | |
| "eval_accuracy": 0.54, | |
| "eval_loss": 0.28571534156799316, | |
| "eval_runtime": 3.1813, | |
| "eval_samples_per_second": 31.434, | |
| "eval_steps_per_second": 4.086, | |
| "step": 650 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 26.0, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 27.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28705817461013794, | |
| "eval_runtime": 3.1784, | |
| "eval_samples_per_second": 31.462, | |
| "eval_steps_per_second": 4.09, | |
| "step": 675 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 27.0, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 28.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2864057421684265, | |
| "eval_runtime": 3.1813, | |
| "eval_samples_per_second": 31.433, | |
| "eval_steps_per_second": 4.086, | |
| "step": 700 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 28.0, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 29.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.286210834980011, | |
| "eval_runtime": 3.1863, | |
| "eval_samples_per_second": 31.385, | |
| "eval_steps_per_second": 4.08, | |
| "step": 725 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 29.0, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2865646183490753, | |
| "eval_runtime": 3.1876, | |
| "eval_samples_per_second": 31.372, | |
| "eval_steps_per_second": 4.078, | |
| "step": 750 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 30.0, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 31.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2837320864200592, | |
| "eval_runtime": 3.1888, | |
| "eval_samples_per_second": 31.359, | |
| "eval_steps_per_second": 4.077, | |
| "step": 775 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 31.0, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 32.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.28590789437294006, | |
| "eval_runtime": 3.1898, | |
| "eval_samples_per_second": 31.35, | |
| "eval_steps_per_second": 4.076, | |
| "step": 800 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 32.0, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 33.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28413906693458557, | |
| "eval_runtime": 3.1938, | |
| "eval_samples_per_second": 31.31, | |
| "eval_steps_per_second": 4.07, | |
| "step": 825 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 33.0, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 34.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2877855598926544, | |
| "eval_runtime": 3.195, | |
| "eval_samples_per_second": 31.299, | |
| "eval_steps_per_second": 4.069, | |
| "step": 850 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 34.0, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 35.0, | |
| "eval_accuracy": 0.61, | |
| "eval_loss": 0.2889203727245331, | |
| "eval_runtime": 3.1944, | |
| "eval_samples_per_second": 31.304, | |
| "eval_steps_per_second": 4.07, | |
| "step": 875 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 35.0, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 36.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2829570770263672, | |
| "eval_runtime": 3.1904, | |
| "eval_samples_per_second": 31.344, | |
| "eval_steps_per_second": 4.075, | |
| "step": 900 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 36.0, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 37.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.28235161304473877, | |
| "eval_runtime": 3.189, | |
| "eval_samples_per_second": 31.358, | |
| "eval_steps_per_second": 4.077, | |
| "step": 925 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 37.0, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 38.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.28012779355049133, | |
| "eval_runtime": 3.1877, | |
| "eval_samples_per_second": 31.37, | |
| "eval_steps_per_second": 4.078, | |
| "step": 950 | |
| }, | |
| { | |
| "best_epoch": 7, | |
| "best_eval_accuracy": 0.63, | |
| "epoch": 38.0, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 39.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.2931237518787384, | |
| "eval_runtime": 3.1852, | |
| "eval_samples_per_second": 31.395, | |
| "eval_steps_per_second": 4.081, | |
| "step": 975 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 39.0, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "learning_rate": 0.01, | |
| "loss": 0.5477, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 40.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.2787815034389496, | |
| "eval_runtime": 3.185, | |
| "eval_samples_per_second": 31.397, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1000 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 40.0, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 41.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2892024517059326, | |
| "eval_runtime": 3.1842, | |
| "eval_samples_per_second": 31.406, | |
| "eval_steps_per_second": 4.083, | |
| "step": 1025 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 41.0, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 42.0, | |
| "eval_accuracy": 0.58, | |
| "eval_loss": 0.2937439978122711, | |
| "eval_runtime": 3.1848, | |
| "eval_samples_per_second": 31.399, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1050 | |
| }, | |
| { | |
| "best_epoch": 38, | |
| "best_eval_accuracy": 0.65, | |
| "epoch": 42.0, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 43.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.28862202167510986, | |
| "eval_runtime": 3.1818, | |
| "eval_samples_per_second": 31.428, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1075 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 43.0, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 44.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.2842145264148712, | |
| "eval_runtime": 3.1861, | |
| "eval_samples_per_second": 31.386, | |
| "eval_steps_per_second": 4.08, | |
| "step": 1100 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 44.0, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 45.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28565531969070435, | |
| "eval_runtime": 3.1841, | |
| "eval_samples_per_second": 31.406, | |
| "eval_steps_per_second": 4.083, | |
| "step": 1125 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 45.0, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 46.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28344452381134033, | |
| "eval_runtime": 3.1839, | |
| "eval_samples_per_second": 31.408, | |
| "eval_steps_per_second": 4.083, | |
| "step": 1150 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 46.0, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 47.0, | |
| "eval_accuracy": 0.56, | |
| "eval_loss": 0.28242892026901245, | |
| "eval_runtime": 3.1842, | |
| "eval_samples_per_second": 31.405, | |
| "eval_steps_per_second": 4.083, | |
| "step": 1175 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 47.0, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 48.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.28662043809890747, | |
| "eval_runtime": 3.1872, | |
| "eval_samples_per_second": 31.376, | |
| "eval_steps_per_second": 4.079, | |
| "step": 1200 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 48.0, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 49.0, | |
| "eval_accuracy": 0.63, | |
| "eval_loss": 0.2801302969455719, | |
| "eval_runtime": 3.1865, | |
| "eval_samples_per_second": 31.382, | |
| "eval_steps_per_second": 4.08, | |
| "step": 1225 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 49.0, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "eval_accuracy": 0.62, | |
| "eval_loss": 0.28511884808540344, | |
| "eval_runtime": 3.1848, | |
| "eval_samples_per_second": 31.399, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1250 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 50.0, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 51.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.2829468846321106, | |
| "eval_runtime": 3.1837, | |
| "eval_samples_per_second": 31.41, | |
| "eval_steps_per_second": 4.083, | |
| "step": 1275 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 51.0, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 52.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.29002857208251953, | |
| "eval_runtime": 3.1829, | |
| "eval_samples_per_second": 31.418, | |
| "eval_steps_per_second": 4.084, | |
| "step": 1300 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 52.0, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 53.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.2781957685947418, | |
| "eval_runtime": 3.1815, | |
| "eval_samples_per_second": 31.431, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1325 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 53.0, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 54.0, | |
| "eval_accuracy": 0.59, | |
| "eval_loss": 0.27934157848358154, | |
| "eval_runtime": 3.1792, | |
| "eval_samples_per_second": 31.454, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1350 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 54.0, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 55.0, | |
| "eval_accuracy": 0.6, | |
| "eval_loss": 0.28086578845977783, | |
| "eval_runtime": 3.1801, | |
| "eval_samples_per_second": 31.446, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1375 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 55.0, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 56.0, | |
| "eval_accuracy": 0.64, | |
| "eval_loss": 0.28146588802337646, | |
| "eval_runtime": 3.1815, | |
| "eval_samples_per_second": 31.432, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1400 | |
| }, | |
| { | |
| "best_epoch": 42, | |
| "best_eval_accuracy": 0.66, | |
| "epoch": 56.0, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 57.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.27978724241256714, | |
| "eval_runtime": 3.182, | |
| "eval_samples_per_second": 31.426, | |
| "eval_steps_per_second": 4.085, | |
| "step": 1425 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 57.0, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 58.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.28314894437789917, | |
| "eval_runtime": 3.1846, | |
| "eval_samples_per_second": 31.401, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1450 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 58.0, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 59.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.27948451042175293, | |
| "eval_runtime": 3.1847, | |
| "eval_samples_per_second": 31.4, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1475 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 59.0, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "learning_rate": 0.005, | |
| "loss": 0.4601, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 60.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2746821343898773, | |
| "eval_runtime": 3.1819, | |
| "eval_samples_per_second": 31.428, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1500 | |
| }, | |
| { | |
| "best_epoch": 56, | |
| "best_eval_accuracy": 0.68, | |
| "epoch": 60.0, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 61.0, | |
| "eval_accuracy": 0.73, | |
| "eval_loss": 0.27249741554260254, | |
| "eval_runtime": 3.1817, | |
| "eval_samples_per_second": 31.429, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1525 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 61.0, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 62.0, | |
| "eval_accuracy": 0.66, | |
| "eval_loss": 0.2840454578399658, | |
| "eval_runtime": 3.182, | |
| "eval_samples_per_second": 31.426, | |
| "eval_steps_per_second": 4.085, | |
| "step": 1550 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 62.0, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 63.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.27392327785491943, | |
| "eval_runtime": 3.1822, | |
| "eval_samples_per_second": 31.425, | |
| "eval_steps_per_second": 4.085, | |
| "step": 1575 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 63.0, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 64.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.27964526414871216, | |
| "eval_runtime": 3.1831, | |
| "eval_samples_per_second": 31.416, | |
| "eval_steps_per_second": 4.084, | |
| "step": 1600 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 64.0, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 65.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.27821072936058044, | |
| "eval_runtime": 3.1837, | |
| "eval_samples_per_second": 31.41, | |
| "eval_steps_per_second": 4.083, | |
| "step": 1625 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 65.0, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 66.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.27570292353630066, | |
| "eval_runtime": 3.1813, | |
| "eval_samples_per_second": 31.433, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1650 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 66.0, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 67.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.27588510513305664, | |
| "eval_runtime": 3.1798, | |
| "eval_samples_per_second": 31.449, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1675 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 67.0, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 68.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.2779083549976349, | |
| "eval_runtime": 3.1791, | |
| "eval_samples_per_second": 31.455, | |
| "eval_steps_per_second": 4.089, | |
| "step": 1700 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 68.0, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 69.0, | |
| "eval_accuracy": 0.67, | |
| "eval_loss": 0.2822496294975281, | |
| "eval_runtime": 3.1788, | |
| "eval_samples_per_second": 31.458, | |
| "eval_steps_per_second": 4.09, | |
| "step": 1725 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 69.0, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 70.0, | |
| "eval_accuracy": 0.65, | |
| "eval_loss": 0.2813234031200409, | |
| "eval_runtime": 3.1815, | |
| "eval_samples_per_second": 31.431, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1750 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 70.0, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 71.0, | |
| "eval_accuracy": 0.68, | |
| "eval_loss": 0.2817613184452057, | |
| "eval_runtime": 3.1828, | |
| "eval_samples_per_second": 31.419, | |
| "eval_steps_per_second": 4.084, | |
| "step": 1775 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 71.0, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 72.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2864615023136139, | |
| "eval_runtime": 3.1825, | |
| "eval_samples_per_second": 31.422, | |
| "eval_steps_per_second": 4.085, | |
| "step": 1800 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 72.0, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 73.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.27701735496520996, | |
| "eval_runtime": 3.1818, | |
| "eval_samples_per_second": 31.429, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1825 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 73.0, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 74.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.28216809034347534, | |
| "eval_runtime": 3.185, | |
| "eval_samples_per_second": 31.397, | |
| "eval_steps_per_second": 4.082, | |
| "step": 1850 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 74.0, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 75.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2782612442970276, | |
| "eval_runtime": 3.1897, | |
| "eval_samples_per_second": 31.351, | |
| "eval_steps_per_second": 4.076, | |
| "step": 1875 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 75.0, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 76.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.2763740122318268, | |
| "eval_runtime": 3.1814, | |
| "eval_samples_per_second": 31.433, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1900 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 76.0, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 77.0, | |
| "eval_accuracy": 0.69, | |
| "eval_loss": 0.2772051692008972, | |
| "eval_runtime": 3.1813, | |
| "eval_samples_per_second": 31.433, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1925 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 77.0, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 78.0, | |
| "eval_accuracy": 0.7, | |
| "eval_loss": 0.27587586641311646, | |
| "eval_runtime": 3.1803, | |
| "eval_samples_per_second": 31.444, | |
| "eval_steps_per_second": 4.088, | |
| "step": 1950 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 78.0, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 79.0, | |
| "eval_accuracy": 0.72, | |
| "eval_loss": 0.27509966492652893, | |
| "eval_runtime": 3.1812, | |
| "eval_samples_per_second": 31.434, | |
| "eval_steps_per_second": 4.086, | |
| "step": 1975 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 79.0, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.4329, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "eval_accuracy": 0.71, | |
| "eval_loss": 0.275318443775177, | |
| "eval_runtime": 3.1812, | |
| "eval_samples_per_second": 31.435, | |
| "eval_steps_per_second": 4.087, | |
| "step": 2000 | |
| }, | |
| { | |
| "best_epoch": 60, | |
| "best_eval_accuracy": 0.73, | |
| "epoch": 80.0, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 80.0, | |
| "step": 2000, | |
| "total_flos": 2.9821702864896e+16, | |
| "train_loss": 0.5270161437988281, | |
| "train_runtime": 1697.455, | |
| "train_samples_per_second": 18.852, | |
| "train_steps_per_second": 1.178 | |
| } | |
| ], | |
| "max_steps": 2000, | |
| "num_train_epochs": 80, | |
| "total_flos": 2.9821702864896e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |